跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.02490 · v2

How Contextual Information in Physical Environments Shapes Decisions with AI

Omar ElgoharyAmanda ElvarsdottirVannessa LeeZhu-Tian Chen
cs.HC
首次提交 2026-11-02 12:05
版本更新 2026-11-05 03:23
雷达观测 2026-11-06 11:06
arXiv ↗PDF ↗

摘要

AI assistants in augmented reality may advise users about parts of the physical world they cannot directly observe. We ask whether contextual information from the situation helps users make better decisions with fallible AI. In an AR visual- search study (N = 32), we varied target prevalence and scene visibility while participants judged whether a key was present, with or without AI assistance. Lowering target prevalence and revealing part of the desk together improved observed AI-assisted accuracy, but our analysis did not demonstrate that this gain came from participants making better use of context. Context also changed which answers seemed plausible: when keys were rare, participants became more skeptical of Present suggestions, but this shift applied to both correct and incorrect suggestions rather than improving selective acceptance of correct advice. Revealing part of the desk also increased decision time, with little evidence of better AI use. Overall, contextual information can shape AI- assisted decisions and increase decision effort without necessarily helping users judge when the AI is correct.

查看 arXiv 原文 ↗阅读 PDF ↗

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存，登录后共享已有结果。

版本差异

仅对比实际观测的元数据，不代表 PDF 正文或实验变化。版本可能不连续；带删除线为移除，下划线为新增。

摘要变化

AI assistants in augmented reality may advise users about parts of the physical world they cannot directly observe. We studiedask whether contextual information from the situation helps peopleusers make better decisions with fallible AI in a partially visible environment. In an AR visual- search study (N = 32), we varied target prevalence and scene visibility while participants judged whether a rendered key was present on a physical desk, with or without AI adviceassistance. We variedLowering target prevalence and scenerevealing visibility.part Focalof analysesthe excludeddesk visibletogether targetsimproved andobserved AI suggestions-assisted thataccuracy, couldbut beour directlyanalysis checked.did Withinnot thisdemonstrate subset,that thethis combinedgain contextcame offrom aparticipants strongermaking priorbetter towarduse absenceof (lowercontext. prevalence)Context andalso achanged target-freewhich halfanswers increasedseemed AI-assistedplausible: accuracywhen bykeys 9.5were percentagerare, points.participants Anbecame exploratorymore analysisskeptical accountedof forPresent mostsuggestions, ofbut this increase throughshift changesapplied into theboth casescorrect and AIincorrect suggestions being evaluated; changed participant responses did not significantlyrather improvethan accuracyimproving beyondselective maintainingacceptance baselineof responsecorrect patternsadvice. The stronger prior reduced agreement with PresentRevealing suggestions,part whileof the target-freedesk halfalso lengthenedincreased decision time. We did not detect more, selectivewith agreementlittle withevidence correctof versusbetter incorrectAI adviceuse. These findingsOverall, motivatecontextual evaluatinginformation situatedcan assistanceshape onAI- matchedassisted casesdecisions and designingincrease waysdecision toeffort obtainwithout evidencenecessarily thathelping canusers resolvejudge claimswhen aboutthe hiddenAI statesis correct.

观测到的版本

相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-05 03:23 · 最新观测版本

How Contextual Information in Physical Environments Shapes Decisions with AI

AI assistants in augmented reality may advise users about parts of the physical world they cannot directly observe. We ask whether contextual information from the situation helps users make better decisions with fallible AI. In an AR visual- search study (N = 32), we varied target prevalence and scene visibility while participants judged whether a key was present, with or without AI assistance. Lowering target prevalence and revealing part of the desk together improved observed AI-assisted accuracy, but our analysis did not demonstrate that this gain came from participants making better use of context. Context also changed which answers seemed plausible: when keys were rare, participants became more skeptical of Present suggestions, but this shift applied to both correct and incorrect suggestions rather than improving selective acceptance of correct advice. Revealing part of the desk also increased decision time, with little evidence of better AI use. Overall, contextual information can shape AI- assisted decisions and increase decision effort without necessarily helping users judge when the AI is correct.

查看此版本解读 →查看此版本原文 ↗
v1 · 2026-11-02 12:05

How Contextual Information in Physical Environments Shapes Decisions with AI

We studied whether contextual information helps people make better decisions with fallible AI in a partially visible environment. In an AR study (N = 32), participants judged whether a rendered key was present on a physical desk, with or without AI advice. We varied target prevalence and scene visibility. Focal analyses excluded visible targets and AI suggestions that could be directly checked. Within this subset, the combined context of a stronger prior toward absence (lower prevalence) and a target-free half increased AI-assisted accuracy by 9.5 percentage points. An exploratory analysis accounted for most of this increase through changes in the cases and AI suggestions being evaluated; changed participant responses did not significantly improve accuracy beyond maintaining baseline response patterns. The stronger prior reduced agreement with Present suggestions, while the target-free half lengthened decision time. We did not detect more selective agreement with correct versus incorrect advice. These findings motivate evaluating situated assistance on matched cases and designing ways to obtain evidence that can resolve claims about hidden states.

查看此版本解读 →查看此版本原文 ↗